We provide additional support for Cowan's claim that short term memory (STM) involves a range of 3–5 tokens, on the basis of language correlational analyses. If language is at least partly learned, linguistic dependency structure should reflect properties of the cognitive components mediating learning; one such component is STM. In this view, the range over which statistical regularity extends in ordinary text would be suggestive of STM span. Our analyses of eight languages are consistent with STM span being about four chunks.

We suggest a neurophysiological account of the short-term memory capacity limit based on a model of visual working memory (Raffone & Wolters, in press). Simulations have revealed a critical capacity limit of about four independent patterns. The model mechanisms may be applicable to working memory in general and they allow a reinterpretation of some of the issues discussed by Cowan.

The capacity of visual attention/STM can be determined by change-detection experiments. Detecting the presence of change leads to an estimate of 4 items, while detecting the absence of change leads to an estimate of 1 item. Thus, there are two magical numbers in vision: 4 and 1. The underlying limits, however, are not necessarily those of central STM.

Cowan makes an intriguing case for a fundamental limit in the number of chunks that can be stored in short term memory (STM). Chunks are collections of concepts that have strong associations to one another and much weaker associations to other chunks. A translation of this definition for the visual domain would be that a visual chunk is a collection of features that belong to the same perceptual group (see also Mahoney & Ullman 1988). Here, we will first address the neuronal mechanisms that may demarcate visual chunks. Then we critically evaluate to what extent these mechanisms might be responsible for the limit on the number of chunks that can be held in STM. We conclude that the clarity with which the psychophysical data point to the number four is not matched by a similarly clear limit imposed by physiological mechanisms.

Cowan argues that the true short-term memory (STM) capacity limit is about 4 items. Functional neuroimaging data converge with this conclusion, indicating distinct neural activity patterns depending on whether or not memory task-demands exceed this limit. STM for verbal information within that capacity invokes focal prefrontal cortical activation that increases with memory load. STM for verbal information exceeding that capacity invokes widespread prefrontal activation in regions associated with executive and attentional processes that may mediate chunking processes to accommodate STM capacity limits.

Cowan defines a chunk as “a collection of concepts that have strong associations to one another and much weaker associations to other chunks currently in use.” This definition does not impose any constraints on the nature and number of elements that can be bound into a chunk. We present an experiment to demonstrate that such limitations exist for visual short-term memory, and that their analysis may lead to important insights into properties of visual memory.

Some of the evidence for a “magical number 4” has come from the study of visual cognition, and Cowan reinterprets such evidence in terms of a single general limit on memory and attention. We evaluate this evidence, including some studies not mentioned by Cowan, and argue that limitations in visual processing are distinct from those involved in other memory phenomena.

Cowan assumes a unitary capacity-limited attentional focus. We argue that two main problems need to be solved before this assumption can complement theoretical knowledge about human cognition. First, it needs to be clarified what exactly the nature of the elements (chunks) within the attentional focus is. Second, an elaborated process model needs to be developed and testable assumptions about the proposed capacity limitation need to be formulated.

The limited capacity for unrelated things is a fact that needs to be explained by a general theory of memory, rather than being itself used as a means of explaining data. A pure storage capacity is therefore not the right assumption for memory research. Instead an explanation is needed of how capacity limitations arise from the interaction between the environment and the cognitive system. The ACT-R architecture, a theory without working memory but a long-term memory based on activation, may provide such an explanation.

Cowan's analysis of human short-term memory (STM) and attention in terms of processing limits in the range of 4 items (or “chunks”) is discussed from the point of view of cognitive neuroscience. Although, Cowan already provides many important theoretical insights, we need to learn more about how to build further bridges between cognitive psychology and cognitive neuroscience.

I like Cowan's review of STM properties and especially his suggestions on the role of attention. I missed, however, a consideration of studies which provide evidence for STM properties in animals. In my commentary, I argue that such evidence can elucidate the biological basis of storage limitations, validating this view by discussing mechanisms which constrain the acquisition of serial information in songbirds.

Cowan has written a meticulous and thought-provoking review of the literature on short-term memory. However, reflections on one area of evidence, that of working memory span, shows the extent to which the research debate can be circumscribed by choice of experimental paradigms.

The aim of our commentary is to strengthen Cowan's proposal for an inherent capacity limitation in STM by suggesting a neurobiological mechanism based on competitive networks and nonlinear oscillations that avoids some of the shortcomings of the scheme discussed in the target article (Lisman & Idiart 1995).

Can we identify individuals with a larger basic capacity than Cowan's proposed limit? Thompson et al. (1993) claimed that Rajan Mahadevan had a basic memory span of 13–15 items. Some of their supporting evidence is reconsidered and additional data are presented from study of another memory expert. More detailed analysis of performance in such cases may yield different conclusions.

A primary proposal of the Cowan target article is that capacity limits arise in working memory because only 4 chunks of information can be attended at one time. This implies a single, unitary attentional focus or resource; we instead propose that relatively independent attentional mech- anisms operate within different cognitive subsystems depending on the demands of the current stimuli and tasks.

Commentators expressed a wide variety of views on whether there is a basic capacity limit of 3 to 5 chunks and, among those who believe in it, about why it occurs. In this response, I conclude that the capacity limit is real and that the concept is strengthened by additional evidence offered by a number of commentators. I consider various arguments why the limit occurs and try to organize these arguments into a conceptual framework or “metatheory” of storage capacity limits meant to be useful in future research to settle the issue. I suggest that principles of memory representation determine what parts of the representation will be most prominent but that limits of attention (or of a memory store that includes only items that have been most recently attended) determine the 3- to 5-chunk capacity limit.

The following questions are considered: Why is it difficult to create a theory of consciousness? What are the contents of consciousness? What kind of theory is acceptable as transparent? and, What is the value of conscious experience?

We suggest that consciousness (C) should be addressed as a multilevel concept. We can provisionally identify at least three, rather than two, levels: Gray's system should relate at least to the lowest of these three levels. Although it is unlikely to be possible to develop a behavioural test for C, it is possible to speculate as to the evolutionary advantages offered by C and how C evolved through succeeding levels. Disturbances in the relationships between the levels of C could underlie mental illness, especially schizophrenia.

Gray's target article presents a model of consciousness that includes several ideas similar to those developed over the past century to explain how sensorimotor information is interpreted by the nervous system. This commentary discusses these ideas and introduces some additional hypotheses, also derived from sensorimotor investigations, that might help improve Gray's model.

Gray's account of a brain mechanism for generating the contents of consciousness is incomplete. Adaptive advantages of conscious functioning need to be sought within the first-person affective sensation motivating flexibly goal-directed actions, as in Humphrey's sensory feedback theory.